Copilot (licence) - Analyze the Copilot Studio agent Tutorial

Unlock the full potential of your assistant with 'Analyze the Copilot Studio Agent.' Dive into powerful analytics that reveal usage statistics, user satisfaction scores, and interaction outcomes. Discover how to enhance performance by monitoring success rates and adapting to user feedback. Don't miss this opportunity to elevate your assistant's effectiveness!

01:25
247 views
Parteger

00:00:08

you can track and test its effectiveness using the available tools.

00:00:13

Click on the agent

00:00:16

Then open activity.

00:00:18

Here you'll find all the conversations conducted

00:00:22

along with useful details such as the name,

00:00:25

the channel used,

00:00:26

and the number of steps.

00:00:30

The evaluation section allows you to run tests with your agent.

00:00:35

To get started,

00:00:36

you can upload a CSV file containing test questions.

00:00:41

You can also use ready-made options

00:00:43

like automatically generating 10 questions using artificial intelligence.

00:00:49

Additional questions can be added manually

00:00:52

or generated again by the AI.

00:00:55

Each test must be given a name to make future analysis easier.

00:01:00

Once everything is set,

00:01:02

simply save the test

00:01:04

or launch it by clicking evaluate.

00:01:08

The test may take several minutes to complete.

00:01:11

When it's finished,

00:01:12

the success rate is displayed.

00:01:15

By clicking on a specific test,

00:01:17

you can access all related discussions

00:01:20

along with their individual results.

00:01:23

This gives you a clearer and more detailed view of each interaction.

00:01:31

If you select show activity map,

00:01:34

you'll be redirected to the activity section of the corresponding conversation.

00:01:41

In the analytics section,

00:01:43

you can explore usage statistics across different time periods.

00:01:47

You'll see every conversation started with the agent,

00:01:51

the user response rate,

00:01:53

and a satisfaction score.

00:01:56

You can also access billing details,

00:01:59

view active users,

00:02:00

and see an estimate of the savings generated by co-pilot.

00:02:05

Conversation outcomes show how sessions handled in classic mode ended.

00:02:11

These outcomes include

00:02:12

resolved,

00:02:13

escalated,

00:02:14

or abandoned.

00:02:16

Tracking the topics associated with these outcomes helps improve the agent.

00:02:24

The use section tracks how many questions the AI answers

00:02:28

and the quality of those responses.

00:02:31

Quality is measured based on completeness,

00:02:34

relevance,

00:02:35

and sources.

00:02:37

Keep in mind metrics require at least 10

00:02:40

answered questions per day to be meaningful.

00:02:46

The satisfaction score helps you identify areas where adjustments may be needed

00:02:51

to encourage greater adoption of the agent.

00:02:55

Thanks to these analysis features,

00:02:57

co-pilot studio provides a comprehensive overview

00:03:01

of your agent's overall effectiveness.

00:03:04

You can clearly understand its performance

00:03:06

and pinpoint opportunities for improvement.

Mandarine AI: WHAT YOU SHOULD KNOW

RELATED COURSES This tutorial is part of the following courses

E-learning Débutant

Vu 406 fois

Get Familiar with Copilot Studio

Build and deploy custom AI assistants in Microsoft Teams with Copilot Studio. Optimize interaction with your teams with a simple interface and features tailored to your collaborative needs.

ALL COURSES